- Email: [email protected]
In-situ monitoring of 1st-order phase transition on InAs(001) surfaces by scanning electron surface microscopy
applied
surface science ELSEVIER
Applied Surface Science 82/83 (1994) 223-227
In-situ monitoring of lst-order phase transition on InAs(OO1) surfaces by scanning electron surface microscopy Hiroshi Yamaguchi
a9*,Yoshikazu Homma b, Yoshiji Horikoshi
a
aNTTBasic Research Laboratories, At@-shi, Kanagawa 243-01, Japan b NTT Interdisciplinary Research Laboratories, Musashino-shi, Tokyo 180, Japan Received 2 May 1994; accepted for publication
8 July 1994
Abstract Domain formation during the phase transition on an InAs(OO1) surface under As pressure was observed directly for the first time by using scanning electron surface microscopy, which is a surface-sensitive scanning electron microscopy technique. Detailed analysis also verified that the role of step edges during transition differs between As-atom desorption and adsorption.
1. Introduction Recent advances in molecular beam epitaxy (MBE) are useful for studying surface elemental processes as well as for fabricating optoelectric devices [l]. Phase transition is commonly observed in many physical systems and is one of the most important physical phenomena observed on MBE-grown semiconductor surfaces. Study of the phase transition provides important information about the surface elemental processes during MBE growth. For elemental semiconductors, like Si and Ge, the surface phase transition is easy to study directly by using in-situ observation techniques like scanning tunneling microscopy @TM) [2] and reflection electron microscopy (REM) [3] because the order-disorder transition occurs in an ultrahigh-vacuum atmosphere without any molecular beam flux. The phase transition on III-V compound semiconductors, however,
* Corresponding
author.
0169-4332/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved SSDI 0169-4332(94)00222-3
is harder to observe because it occurs reversibly only under Group-V pressure [4,5]. Studies of this phase transition have therefore until now been based on reflection high-energy electron diffraction (RHEED) observations. In the work reported here, however, we directly observed domain formation during the phase transition on an InAs(OO1) surface under As pressure by using scanning electron surface microscopy [6,7], which is a surface-sensitive ultrahigh-vacuum scanning electron microscopy (UHV-SEM) technique. Comparing these experimental results with the results of Monte Carlo simulation confirms the strong attractive lateral interaction between surface species on InA.4001) surfaces.
2. lst-order surfaces
phase
transition
on
IuAs(OO1)
An InAs(OO1) surface shows a lst-order phase transition between As-stabilized (2 X 4) and Instabilized (4 X 2) surfaces: as the substrate tempera-
ture changes under As pressure. the electron retlcctivity changes discontinuously with hysteresis hetween these two surfaces [5]. Monte Carlo simulation with a two-dimensional lattice-gas model well explains the phenomena [8,o] and indicates that the strong lateral interaction on the surface induces the formation of domains for each surface structure [o]. In the simulation, we take two kinds on interaction into account. One, with an energy of E,, is a vertical interaction between surface As atoms and underlying In atoms. The other, with an energy of E,, is a lateral interaction between surface As atoms. No activation barrier was used for As adsorption for simplicity. The simulation shows that the lst-order transition occurs with an E, greater than 1.76kT,,, where k is the Boltzmann constant and T,, is the temperature of transition from the (2 X 41 structure to the (4 X 21 structure. The discontinuous transition between these two structures on InAs(OOl), therefore, seems to be caused by strong lateral interaction between surface species. Fig. 1 shows a surface stoichiometry change obtained from the simulations. A rectangular lattice with the size of 72 X 72 was used. The white square corresponds to the As-covered site and the black one to the In-exposed site. We used E, and E, values of 2.16 and 0.14 eV, respectively, which reproduce the As-pressure dependence of the transition temperature and hysteresis width [lo]. Because of the strong lateral interaction between the surface As units, As atoms desorbed from the surface by forming domains. Scanning tunneling microscopy observation of heated InAs surfaces under ultrahigh vacuum showed that the (4 X 2) domains were formed in the (2 X 4) structure [I I]. This result suggests that the transition occurs by forming domains for InAs surface even in the thermal equilibrium condition under As pressure. Our UHV-SEM system directly confirmed that the domain formation predicted by Monte Carlo simulation occurs under the same thermal equilibrium condition as in RHEED observation.
3. Experimental
procedure
The details of our experimental set-up have been already reported [6,7]. A UHV-SEM instrument equipped with a field-emission electron gun and
(a) t = 0 s
(c) t = 60 s
Fig. 1. Simulated surface structures during the phase transition lnAs(001) surfaces at 30 5 intervals.
on
liquid-nitrogen-cooled shroud was used for surface imaging. The background pressure during SEM observation was I X IW” Torr. A Knudsen effusion cell for elemental As supplied an As flux of about 1 X lo-” Torr. which was monitored by a flux gauge monitor placed at the sample. The specimen stage could be tilted from 0” (grazing incidence) to
H. Yamaguchi et al. /Applied Surface Science 82 /83
yivy
(1994) 223-227
225
contrast corresponds to a (4 X 2) structure, and the growth of (4 X 2) domains is clearly visible. These results directly indicate that domains are formed
e3cl” detector i) As flux Fig. 2. Schematic incidence,
illustration
of electron
beam
and As beam
90” (normal incidence). In our in-situ observation of the phase transition, the electron beam was incident from the top at 30” to the surface, and the As flux was incident from the bottom at 26” to the sample surface (Fig. 2). The electron beam energy was 25 keV and the best resolution was about 5 nm. Specimens measuring 7 X 5 mm* were cut from an n-type InAs(OO1) wafer misoriented by 1” toward the [l lo] direction. Specimens were etched in a H,SO, : H,O, : H,O = 20 : 1: 1 solution. A 200 nm thick InAs buffer layer was grown at 480°C in an MBE chamber before the observations, and a protective layer of elemental As was formed by depositing As, on the surface cooled to - 10°C. The sample was then mounted, using indium, on Si substrate that acted as a resistive heater and transferred, through air, to the UHV-SEM chamber where the protective layer was removed by heating the sample at 300°C. The SEM observation was performed while changing the sample temperature at about 2”C/min under an As pressure of about 1 X 10e6 Torr.
4. Experimental
results and discussion
Fig. 3 shows the successively obtained secondary electron images of an InAs(OO1) surface at 480°C. The 10 s interval was necessary for scanning the area. The RHEED pattern showed that the surface structure changed from (2 X 4) to (4 X 2) at this temperature. The domain with dark contrast thus corresponds to a (2 X 4) structure, that with bright
Fig. 3. Secondary electron images of an InAs(OO1) surface during the transition from a (2 X 4) to a (4 X 2.) surface at 10 s ((a), (b)) and 20 s ((b), (c)j intervals. The image size is 2.3 Frn (horizontal) X 2.9 pm (vertical).
726
H. Yamaguchi el al. /Applied Surface Science 82 /X3 (1994) 223-227
during the transition as predicted by the two-dimensional lattice-gas model [9]. The expected terrace size along the [IlO] direction, calculated from the misorientation angle of l”, is 16 nm. The size actually observed, however, in Fig. 3, for example, is about 50 nm, showing step bunching by forming a multi-step structure. The step bunching has also been noted by STM observation [12] and SEM observation [7] of growth-interrupted GaAs(001) surfaces. The formed (4 X 2) domain, when about half the coverage of the surface changed to (4 X 2), has an average size in the range 20-100 nm in the [llO] direction and 100-500 nm in the [liO] direction. This anisotropic shape of the domains is due to them being bounded by the multisteps running in the [liO] direction. Fig. 4 shows the successively obtained secondary electron images of an InAs(OO1) surface at about 470°C when the sample was cooled and the transition was from (4 X 2) to (2 X 4). This transition temperature, 10°C lower than that from (2 X 4) to (4 X 21, shows the same 10°C hysteresis with that obtained by RHEED observations [5]. The role of monomolecular steps during the transition from (4 X 2) to (2 X 4) differs from that from (2 X 4) to (4 X 2). At the beginning of the transition from (4 X 2), the steps have darker contrast than the terraces shown in Fig. 4a. This enhanced contrast at the multi-steps indicates that the As atoms preferentially adsorb at the step edge. This was not observed for the transition from (2 X 4) to (4 X 2). The steps in Fig. 3a do not have a brighter contrast than the terraces, but have a rather darker contrast, indicating that the step edge does not act as the desorption site. Growth of (2 X 4) domains from the step edge was observed when the temperature was decreased for 2 or 3°C below the temperature at which the contrast enhancement was observed (Fig. 4b). Because the As atoms do not desorb from the step edge, the As atoms at the higher side of the step edge (A in Fig. 5a) must be as energetically stable as those on the terraces. If the As atoms have no lateral interaction across the step edge and a vertical interaction with the underlying In atoms comparable to those on the terraces, the As desorption would begin at the step edge [8]. The experimental result shows that this is not the case. The adsorption, on the other hand, begins from the step edge as would be ex-
Fig. 4. Secondary electron image of an InAs(OO1) surface during the transition from a (4 X 2) to a (2 X 4) surface; (a) was obtained just after the transition began, and (b) was obtained 50 s later. The image size is 2.3 @rn (horizontal) X 2.9 pm (vertical).
petted from the lattice-gas model. The role of step edges during the transition, therefore, differs between As-atom desorption and adsorption. We can explain this difference by taking into account the surface atomic structures. The As atoms at the higher side of the step edge (A in Fig. 5a) can be more stable than expected from the lattice-gas model, because (2 X 4) dimer-vacancy row structures [13] are formed. STM observation of the InAs surface of a misoriented surface toward the [110] direction shows that the step edge is in the dimer-vacancy row, and the terrace-size increase is in the unit of 4 X periodicity [lo]. This suggests that the step edge can become more stable by forming dimer-vacancy row structures. For example, the As atom A in Fig. Sa probably has almost the same stability with B. For
H. Yamaguchi et al. /Applied Surface Science 82/83 (a) As
covered
(b) In covered
(2x4)
(1994) 223-227
227
covered (2 X 4) and In-covered (4 X 2) structures on an InAs(OO1) surface under As pressure clearly indicates the domain formation expected from Monte Carlo simulation with a two-dimensional lattice-gas model. The observation also verifies that the role of step edges differs between As-atom desorption and adsorption during the transition. The step edge does not play an important role in desorption but does in adsorption. This difference can be explained by taking into account the surface atomic structures.
(4x2)
Acknowledgement 0
Asatom
0
@
As dangling
bond
0
In dangling
bond
In atom
Fig. 5. Schematic side view of the atomical arrangement for (a) As-covered (2X4) and (b) In-covered (4 X 2) surfaces with a monomolecular step. A dimer-vacancy row structure proposed by Chadi [13] and an In dimer structure are assumed to be formed on (a) and (b), respectively.
As adsorption, however, the initial surface reconstruction is (4 X 2). Although the detailed atomic structure for the (4 X 2) surface is still under discussion [14], In dangling bonds (C in Fig. 5b) exist at the step edge. An As atom at E in Fig. 5 has to break two In dimers, G and H, to form stable chemical bonds with In atoms in contrast to that at D, which has to break only one In dimer F to form them. Therefore, an As atom is thought to be more easily adsorbed on D than E in Fig. 5b. The model well explains the role of step edges in desorption and adsorption of As atoms for a surface misoriented toward the [110] direction. Different roles of the step edge will be observed with a surface misoriented toward the [liO] direction since the direction of the dangling bond and the shape of the step edge is different as reported for the growth process during migration-enhanced epitaxy on misoriented surfaces
Ml. 5. Conclusions The use of scanning electron surface microscopy to directly observe the phase transition between As-
We thank Dr. Tatsuya Kimura for his continuous encouragement of this work.
References [l] See, for example, Molecular Beam Epitaxy and Heterostructures, Eds. L.L. Chang and K. Ploog (Nijhoff, Dordrecht, 1985). [2] R.M. Feenstra, A.J. Slavin, GA. Held and M.A. Lutz, Phys. Rev. Len. 66 (1991) 32.57; S. Kitamura, T. Sato and M. Iwatsuki, Nature 351 (1991) 215. [3] N. Osakabe, Y. Tanishiro, K. Yagi and G. Honjo, Surf. Sci. 109 (1981) 353; W. Telieps and E. Bauer, Surf. Sci. 162 (1985) 163 [4] A.Y. Cho, J. Appl. Phys. 42 (1971) 2074 [5] H. Yamaguchi and Y. Horikoshi, Phys. Rev. B 45 (1992) 1511. [6] Y. Homma, M. Tomita and T. Hayashi, Ultramicroscopy 52 (1993) 187. 171 Y. Homma, J. Osaka and N. Inoue, Jpn. J. Appl. Phys. 33 (1994) L563. [8] H. Yamaguchi and Y. Horikoshi, Phys. Rev. L&t. 70 (1993) 1299. [9] H. Yamaguchi and Y. Horikoshi, J. Cryst. Growth 127 (1993) 976. [lo] H. Yamaguchi and Y. Horikoshi, Appl. Phys. Lett. 64 (1994), to be published. [Ill H. Yamaguchi and Y. Horikoshi, Jpn. J. Appl. Phys. 33 (1994) 716. [12] T. Ide, A. Yamashita and T. Mizutani, Phys. Rev. B 46 (1992) 1905. 1131 D.J. Chadi, J. Vat. Sci. Technol. A 5 (1987) 834. [14] S. Ohkouchi and I. Tanaka, Appl. Phys. Len. 59 (1991) 1588. [1.5] Y. Horikoshi, H. Yamaguchi, F. Briones and M. Kawashima, J. Cryst. Growth 105 (1990) 326.
surface science ELSEVIER
Applied Surface Science 82/83 (1994) 223-227
In-situ monitoring of lst-order phase transition on InAs(OO1) surfaces by scanning electron surface microscopy Hiroshi Yamaguchi
a9*,Yoshikazu Homma b, Yoshiji Horikoshi
a
aNTTBasic Research Laboratories, At@-shi, Kanagawa 243-01, Japan b NTT Interdisciplinary Research Laboratories, Musashino-shi, Tokyo 180, Japan Received 2 May 1994; accepted for publication
8 July 1994
Abstract Domain formation during the phase transition on an InAs(OO1) surface under As pressure was observed directly for the first time by using scanning electron surface microscopy, which is a surface-sensitive scanning electron microscopy technique. Detailed analysis also verified that the role of step edges during transition differs between As-atom desorption and adsorption.
1. Introduction Recent advances in molecular beam epitaxy (MBE) are useful for studying surface elemental processes as well as for fabricating optoelectric devices [l]. Phase transition is commonly observed in many physical systems and is one of the most important physical phenomena observed on MBE-grown semiconductor surfaces. Study of the phase transition provides important information about the surface elemental processes during MBE growth. For elemental semiconductors, like Si and Ge, the surface phase transition is easy to study directly by using in-situ observation techniques like scanning tunneling microscopy @TM) [2] and reflection electron microscopy (REM) [3] because the order-disorder transition occurs in an ultrahigh-vacuum atmosphere without any molecular beam flux. The phase transition on III-V compound semiconductors, however,
* Corresponding
author.
0169-4332/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved SSDI 0169-4332(94)00222-3
is harder to observe because it occurs reversibly only under Group-V pressure [4,5]. Studies of this phase transition have therefore until now been based on reflection high-energy electron diffraction (RHEED) observations. In the work reported here, however, we directly observed domain formation during the phase transition on an InAs(OO1) surface under As pressure by using scanning electron surface microscopy [6,7], which is a surface-sensitive ultrahigh-vacuum scanning electron microscopy (UHV-SEM) technique. Comparing these experimental results with the results of Monte Carlo simulation confirms the strong attractive lateral interaction between surface species on InA.4001) surfaces.
2. lst-order surfaces
phase
transition
on
IuAs(OO1)
An InAs(OO1) surface shows a lst-order phase transition between As-stabilized (2 X 4) and Instabilized (4 X 2) surfaces: as the substrate tempera-
ture changes under As pressure. the electron retlcctivity changes discontinuously with hysteresis hetween these two surfaces [5]. Monte Carlo simulation with a two-dimensional lattice-gas model well explains the phenomena [8,o] and indicates that the strong lateral interaction on the surface induces the formation of domains for each surface structure [o]. In the simulation, we take two kinds on interaction into account. One, with an energy of E,, is a vertical interaction between surface As atoms and underlying In atoms. The other, with an energy of E,, is a lateral interaction between surface As atoms. No activation barrier was used for As adsorption for simplicity. The simulation shows that the lst-order transition occurs with an E, greater than 1.76kT,,, where k is the Boltzmann constant and T,, is the temperature of transition from the (2 X 41 structure to the (4 X 21 structure. The discontinuous transition between these two structures on InAs(OOl), therefore, seems to be caused by strong lateral interaction between surface species. Fig. 1 shows a surface stoichiometry change obtained from the simulations. A rectangular lattice with the size of 72 X 72 was used. The white square corresponds to the As-covered site and the black one to the In-exposed site. We used E, and E, values of 2.16 and 0.14 eV, respectively, which reproduce the As-pressure dependence of the transition temperature and hysteresis width [lo]. Because of the strong lateral interaction between the surface As units, As atoms desorbed from the surface by forming domains. Scanning tunneling microscopy observation of heated InAs surfaces under ultrahigh vacuum showed that the (4 X 2) domains were formed in the (2 X 4) structure [I I]. This result suggests that the transition occurs by forming domains for InAs surface even in the thermal equilibrium condition under As pressure. Our UHV-SEM system directly confirmed that the domain formation predicted by Monte Carlo simulation occurs under the same thermal equilibrium condition as in RHEED observation.
3. Experimental
procedure
The details of our experimental set-up have been already reported [6,7]. A UHV-SEM instrument equipped with a field-emission electron gun and
(a) t = 0 s
(c) t = 60 s
Fig. 1. Simulated surface structures during the phase transition lnAs(001) surfaces at 30 5 intervals.
on
liquid-nitrogen-cooled shroud was used for surface imaging. The background pressure during SEM observation was I X IW” Torr. A Knudsen effusion cell for elemental As supplied an As flux of about 1 X lo-” Torr. which was monitored by a flux gauge monitor placed at the sample. The specimen stage could be tilted from 0” (grazing incidence) to
H. Yamaguchi et al. /Applied Surface Science 82 /83
yivy
(1994) 223-227
225
contrast corresponds to a (4 X 2) structure, and the growth of (4 X 2) domains is clearly visible. These results directly indicate that domains are formed
e3cl” detector i) As flux Fig. 2. Schematic incidence,
illustration
of electron
beam
and As beam
90” (normal incidence). In our in-situ observation of the phase transition, the electron beam was incident from the top at 30” to the surface, and the As flux was incident from the bottom at 26” to the sample surface (Fig. 2). The electron beam energy was 25 keV and the best resolution was about 5 nm. Specimens measuring 7 X 5 mm* were cut from an n-type InAs(OO1) wafer misoriented by 1” toward the [l lo] direction. Specimens were etched in a H,SO, : H,O, : H,O = 20 : 1: 1 solution. A 200 nm thick InAs buffer layer was grown at 480°C in an MBE chamber before the observations, and a protective layer of elemental As was formed by depositing As, on the surface cooled to - 10°C. The sample was then mounted, using indium, on Si substrate that acted as a resistive heater and transferred, through air, to the UHV-SEM chamber where the protective layer was removed by heating the sample at 300°C. The SEM observation was performed while changing the sample temperature at about 2”C/min under an As pressure of about 1 X 10e6 Torr.
4. Experimental
results and discussion
Fig. 3 shows the successively obtained secondary electron images of an InAs(OO1) surface at 480°C. The 10 s interval was necessary for scanning the area. The RHEED pattern showed that the surface structure changed from (2 X 4) to (4 X 2) at this temperature. The domain with dark contrast thus corresponds to a (2 X 4) structure, that with bright
Fig. 3. Secondary electron images of an InAs(OO1) surface during the transition from a (2 X 4) to a (4 X 2.) surface at 10 s ((a), (b)) and 20 s ((b), (c)j intervals. The image size is 2.3 Frn (horizontal) X 2.9 pm (vertical).
726
H. Yamaguchi el al. /Applied Surface Science 82 /X3 (1994) 223-227
during the transition as predicted by the two-dimensional lattice-gas model [9]. The expected terrace size along the [IlO] direction, calculated from the misorientation angle of l”, is 16 nm. The size actually observed, however, in Fig. 3, for example, is about 50 nm, showing step bunching by forming a multi-step structure. The step bunching has also been noted by STM observation [12] and SEM observation [7] of growth-interrupted GaAs(001) surfaces. The formed (4 X 2) domain, when about half the coverage of the surface changed to (4 X 2), has an average size in the range 20-100 nm in the [llO] direction and 100-500 nm in the [liO] direction. This anisotropic shape of the domains is due to them being bounded by the multisteps running in the [liO] direction. Fig. 4 shows the successively obtained secondary electron images of an InAs(OO1) surface at about 470°C when the sample was cooled and the transition was from (4 X 2) to (2 X 4). This transition temperature, 10°C lower than that from (2 X 4) to (4 X 21, shows the same 10°C hysteresis with that obtained by RHEED observations [5]. The role of monomolecular steps during the transition from (4 X 2) to (2 X 4) differs from that from (2 X 4) to (4 X 2). At the beginning of the transition from (4 X 2), the steps have darker contrast than the terraces shown in Fig. 4a. This enhanced contrast at the multi-steps indicates that the As atoms preferentially adsorb at the step edge. This was not observed for the transition from (2 X 4) to (4 X 2). The steps in Fig. 3a do not have a brighter contrast than the terraces, but have a rather darker contrast, indicating that the step edge does not act as the desorption site. Growth of (2 X 4) domains from the step edge was observed when the temperature was decreased for 2 or 3°C below the temperature at which the contrast enhancement was observed (Fig. 4b). Because the As atoms do not desorb from the step edge, the As atoms at the higher side of the step edge (A in Fig. 5a) must be as energetically stable as those on the terraces. If the As atoms have no lateral interaction across the step edge and a vertical interaction with the underlying In atoms comparable to those on the terraces, the As desorption would begin at the step edge [8]. The experimental result shows that this is not the case. The adsorption, on the other hand, begins from the step edge as would be ex-
Fig. 4. Secondary electron image of an InAs(OO1) surface during the transition from a (4 X 2) to a (2 X 4) surface; (a) was obtained just after the transition began, and (b) was obtained 50 s later. The image size is 2.3 @rn (horizontal) X 2.9 pm (vertical).
petted from the lattice-gas model. The role of step edges during the transition, therefore, differs between As-atom desorption and adsorption. We can explain this difference by taking into account the surface atomic structures. The As atoms at the higher side of the step edge (A in Fig. 5a) can be more stable than expected from the lattice-gas model, because (2 X 4) dimer-vacancy row structures [13] are formed. STM observation of the InAs surface of a misoriented surface toward the [110] direction shows that the step edge is in the dimer-vacancy row, and the terrace-size increase is in the unit of 4 X periodicity [lo]. This suggests that the step edge can become more stable by forming dimer-vacancy row structures. For example, the As atom A in Fig. Sa probably has almost the same stability with B. For
H. Yamaguchi et al. /Applied Surface Science 82/83 (a) As
covered
(b) In covered
(2x4)
(1994) 223-227
227
covered (2 X 4) and In-covered (4 X 2) structures on an InAs(OO1) surface under As pressure clearly indicates the domain formation expected from Monte Carlo simulation with a two-dimensional lattice-gas model. The observation also verifies that the role of step edges differs between As-atom desorption and adsorption during the transition. The step edge does not play an important role in desorption but does in adsorption. This difference can be explained by taking into account the surface atomic structures.
(4x2)
Acknowledgement 0
Asatom
0
@
As dangling
bond
0
In dangling
bond
In atom
Fig. 5. Schematic side view of the atomical arrangement for (a) As-covered (2X4) and (b) In-covered (4 X 2) surfaces with a monomolecular step. A dimer-vacancy row structure proposed by Chadi [13] and an In dimer structure are assumed to be formed on (a) and (b), respectively.
As adsorption, however, the initial surface reconstruction is (4 X 2). Although the detailed atomic structure for the (4 X 2) surface is still under discussion [14], In dangling bonds (C in Fig. 5b) exist at the step edge. An As atom at E in Fig. 5 has to break two In dimers, G and H, to form stable chemical bonds with In atoms in contrast to that at D, which has to break only one In dimer F to form them. Therefore, an As atom is thought to be more easily adsorbed on D than E in Fig. 5b. The model well explains the role of step edges in desorption and adsorption of As atoms for a surface misoriented toward the [110] direction. Different roles of the step edge will be observed with a surface misoriented toward the [liO] direction since the direction of the dangling bond and the shape of the step edge is different as reported for the growth process during migration-enhanced epitaxy on misoriented surfaces
Ml. 5. Conclusions The use of scanning electron surface microscopy to directly observe the phase transition between As-
We thank Dr. Tatsuya Kimura for his continuous encouragement of this work.
References [l] See, for example, Molecular Beam Epitaxy and Heterostructures, Eds. L.L. Chang and K. Ploog (Nijhoff, Dordrecht, 1985). [2] R.M. Feenstra, A.J. Slavin, GA. Held and M.A. Lutz, Phys. Rev. Len. 66 (1991) 32.57; S. Kitamura, T. Sato and M. Iwatsuki, Nature 351 (1991) 215. [3] N. Osakabe, Y. Tanishiro, K. Yagi and G. Honjo, Surf. Sci. 109 (1981) 353; W. Telieps and E. Bauer, Surf. Sci. 162 (1985) 163 [4] A.Y. Cho, J. Appl. Phys. 42 (1971) 2074 [5] H. Yamaguchi and Y. Horikoshi, Phys. Rev. B 45 (1992) 1511. [6] Y. Homma, M. Tomita and T. Hayashi, Ultramicroscopy 52 (1993) 187. 171 Y. Homma, J. Osaka and N. Inoue, Jpn. J. Appl. Phys. 33 (1994) L563. [8] H. Yamaguchi and Y. Horikoshi, Phys. Rev. L&t. 70 (1993) 1299. [9] H. Yamaguchi and Y. Horikoshi, J. Cryst. Growth 127 (1993) 976. [lo] H. Yamaguchi and Y. Horikoshi, Appl. Phys. Lett. 64 (1994), to be published. [Ill H. Yamaguchi and Y. Horikoshi, Jpn. J. Appl. Phys. 33 (1994) 716. [12] T. Ide, A. Yamashita and T. Mizutani, Phys. Rev. B 46 (1992) 1905. 1131 D.J. Chadi, J. Vat. Sci. Technol. A 5 (1987) 834. [14] S. Ohkouchi and I. Tanaka, Appl. Phys. Len. 59 (1991) 1588. [1.5] Y. Horikoshi, H. Yamaguchi, F. Briones and M. Kawashima, J. Cryst. Growth 105 (1990) 326.